Current pulse generators are typically used in telephone or telecommunications applications, for example, in a system for communication between sensors and actuators which uses the (unshielded) power supply line for data transmission and operates in an industrial environment where electromagnetic interference is present and there are problems due to the nature of the transmission medium itself. These problems can cause data loss and particular care must therefore be taken regarding the waveform of the signal used to transmit the data, particularly regarding the times for transition from one state to the other.
Many factors can influence data transmission between active nodes or can interfere with nearby equipment. For example, the nature of the transmission medium is one of the factors that can influence data transmission. When a waveform travels along any communication channel, it loses part of its energy due to interaction between the waveform and the structure of the channel itself. This energy loss leads to an attenuation of the low- and high-frequency harmonics, with a consequent distortion of the signal during reception. It is therefore necessary to use waveforms whose harmonic components lie within the frequency band of the channel. Another negative factor is electromagnetic interference. The line over which current pulses travel is a source of electromagnetic forces which radiate energy (since the line is unshielded) into the surrounding environment, interfering with nearby equipment. A limit to this phenomenon is achieved by limiting the frequency band of the transmitted signal, limiting its switching times.
Another interference factor is noise induced externally. Due to the switching of electric circuit breakers, to the starting of neon lamps, electric motors, etc., additional and local voltages may be generated which by electromagnetic coupling to the line can cause overcurrents which are added to the transmitted signal. Bits which alter the content of the transmitted message are therefore added to the received signal. This effect is minimized by using, for each bit to be transmitted, two waveforms with opposite phases which are sent simultaneously over both wires of the line. In this manner the advantage of keeping the current absorbed by the line during transmissions constant is also achieved.
When a plurality of devices arranged physically distant from each other are used in the same system, these devices can have different electrical performance due both to the variations in the manufacturing process and to the difference in the operating temperatures to which they are subjected. These variations affect the switching times of the pulse and can therefore cause errors.
During reception, the acquired signal is sampled at a frequency which is higher than the transmission frequency and is then processed so as to extract the information contained therein. A change in the switching times results in a variation in the duration of the pulse and therefore a different reading of the number of samples. If this difference becomes large, the information associated with the pulse may be lost. The transmission protocol sets the maximum value of the error that can occur during reception to a value which is a fraction, e.g. a few percent, of the transmission frequency.
Therefore, the waveform of the current that flows over the two lines is desirably of the type shown in FIG. 1, where I+ and I- are currents flowing on the two lines, Idc is the average current that flows over the line, and T, T.sub.s and T.sub.d are the duration and the switching times of the pulse. One circuit solution capable of generating a pulse of the type shown in FIG. 1 is shown in FIG. 2, where the signal to be transmitted is applied as a logic state to the inputs Vdiff of a differential pair Q1, Q2 provided by two bipolar transistors.
The current I+ that flows over the line at the voltage V+ is provided by the collector current Ic1, while the current I- that flows over the line at the voltage V- is provided by the collector current Ic2, mirrored by two PNP transistors Q3 and Q4. The inductors L1 and L2 are meant to prevent the signal from returning toward the power supply Vdc, closing instead on a capacitor C arranged on the receiver of the successive node. The biasing current IL is chosen so as to be constant, using a constant-current source external to the device.
The core of the circuit is provided by the differential pair Q1 and Q2. It is known that the collector currents of the two transistors Q1 and Q2 depend on the value of the driving voltage Vdiff. For voltages in the interval +/- 26 mV, their behavior is linear with respect to the input voltage. ##EQU1##
where Vdiff is the driving voltage, IL is the biasing current and Vt is the thermal voltage.
The duration of the transition times T(s,d) (with reference to FIG. 1) of the collector currents is obtained by differentiating the currents Ic1 and Ic2 with respect to time: ##EQU2##
where .DELTA.Ic1=.DELTA.Ic2=.DELTA.Ic and .DELTA.(Vdiff)/.DELTA.T is the variation in the control voltage as a function of time. By defining the switching times T(s,d) as the rise and fall times of the currents Ic1 and Ic2, one obtains, by rewriting the above equation in terms of T(s,d): ##EQU3##
Assuming .DELTA.(Vdiff)/.DELTA.T to be constant, the term T(s,d) is a function of the term Vt, which depends in a linear fashion on the temperature according to the rule Vt=KT/q, and therefore:
Ts,d(T)=.alpha.Vt(T)
For temperature variations in the range between -40.degree. and +125.degree., the resulting absolute variation in switching times is unacceptable. If the error on the currents introduced by the difference between the transistors Q1 and Q2 is added to this, the switching time variation deteriorates even further.
An improvement to the circuit of FIG. 2 is achieved, as shown in FIG. 3, by adding degeneration resistors in series to the emitters of the differential stage formed by the transistors Q1 and Q2. In this manner, by adding resistors R, the input dynamic range is increased and at the same time both Vt and the error due to the difference between the transistors Q1 and Q2 becomes negligible. FIG. 3 also includes the circuit for driving the differential stage, which is formed by a capacitor C1 and by a current source I=I* which are driven in a push-pull configuration by the voltage Vin. In this case, the voltage variation D(Vdiff)DT depends on the charging and discharging of the capacitor C1 at a constant current I according to the relation: ##EQU4##
With these constraints, the switching times T(s,d) with respect to the circuit of FIG. 3 are: ##EQU5##
where A is a coefficient and R is the value of the degeneration resistor which is series-connected to the emitter terminals of the transistors Q1 and Q2.
The dynamic range of the voltage Vdiff=V1-V2 is +/-R*IL, approximately 1 volt, so that the term Vt (26 mV) becomes negligible; moreover, the offset of the differential stage is due to the degeneration resistors R, which can be produced with a higher accuracy than the pair of transistors Q1 and Q2. However, also with this solution, the switching times are not perfectly controlled and independent of process and temperature variations.